Sign In

Communications of the ACM

ACM TechNews

Annealing Processor Excels at Solving Combinatorial Optimization Problems


View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
The new processor.

Researchers in Japan have developed a new annealer architecture to handle combinatorial optimization problems.

Credit: Tokyo Institute of Technology

Researchers at Japan's Tokyo Institute of Technology, Hokkaido University, the University of Tokyo, and Hitachi have developed a new annealer architecture to handle combinatorial optimization problems.

The new processor, called STATICA (Stochastic Cellular Automata Annealer Architecture), is designed to address challenges such as portfolio, logistic, and traffic flow optimization when they are expressed in the form of Ising models.

Other annealers calculate spin switchings one by one, making it a serial process. With STATICA, that updating is performed in parallel using stochastic cellular automata (SCA)—a means of simulating complex systems using the interactions of a large number of neighboring "cells" with simple updating rules and some randomness.

SCA introduces copies of the original spins into the process, and all original spin-spin interactions are redirected to their individual replica spins.

From IEEE Spectrum
View Full Article

 

Abstracts Copyright © 2020 SmithBucklin, Washington, DC, USA


 

No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account
ACM Resources